JP2015036923A - Assessment collection device, assessment ranking preparation device, method for assessment collection and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an assessment collection device, an assessment ranking preparation device, a method for assessment collection and a program that enable a collection of an assessment having high reliability about a product or service.SOLUTION: An assessment collection device collecting a numerical figure pertaining to an assessment of a product or service on the basis of documents collected from Web sites comprises: a collection unit 30 that collects documents pertaining to assessments of products or services written in the Web pages having names of the products or services described; a division part 40 that divides the document collected by the collection part 30 into words; a storage unit 20 that preliminarily stores words matching or similar to affection expression words, and numerical figures corresponding to the words by associating the words with the numerical figures, respectively; and a collection unit 70 that, when the word matching or similar to the affection expression word is included in the word divided by the division unit 40 on the basis of a content of the storage unit 20, collects the numerical figure corresponding to the word.

Description

  The present application relates to an evaluation totaling apparatus, an evaluation rank creating apparatus, an evaluation totaling method, and a program for totalizing evaluations of products or services.

  There is a service that provides a recommended movie related to one movie based on emotion items associated with numerical values (see, for example, Patent Document 1). The service according to Patent Document 1 inputs emotion items included in work evaluation data provided by a third party, totals numerical values corresponding to the input emotion items, and displays the total result.

JP 2005-115877 A

  However, in the case of the service according to Patent Document 1, the number of pieces of work evaluation data provided by a third party and the number of emotion items included in the piece evaluation data are limited, and the trust regarding the total number of recommended movies is limited. Poor nature.

  The present application has been made in view of such circumstances. An object of the present invention is to provide an evaluation totaling apparatus, an evaluation rank creating apparatus, an evaluation totaling method, and a program capable of totaling highly reliable evaluations regarding products or services.

  One aspect of the present application is an evaluation and aggregation device that aggregates numerical values related to evaluation of goods or services based on documents collected from a website, and is written on a web page on which the names of goods or services are written. A collection unit that collects documents related to product or service evaluation from a website, a division unit that divides the documents collected by the collection unit into words, and a phrase that matches and resembles emotional phrases and corresponding to the phrases respectively. Corresponding to the phrase when the phrase divided by the dividing unit includes a phrase that matches or is similar to the emotion expression phrase based on the contents of the storage unit that stores each numerical value in advance in association with each other And a counting unit that counts the numerical values.

  According to one aspect of the present application, highly reliable evaluations regarding products or services can be aggregated.

It is a block diagram which shows the hardware structural example of an evaluation order creation apparatus. It is explanatory drawing which shows an example of the record layout of a merchandise service table. It is explanatory drawing which shows an example of the record layout of a review table. It is explanatory drawing which shows an example of the record layout of a search keyword table. It is explanatory drawing which shows an example of the record layout of a connection word / phrase table. It is explanatory drawing which shows an example of the record layout of the score table classified by review. It is explanatory drawing which shows an example of the record layout of a ranking table. It is explanatory drawing which shows an example of the content of the dependency relation weight value file. It is a functional block diagram which shows the function structural example of an evaluation order creation apparatus. It is a flowchart which shows an example of the procedure of an evaluation ranking preparation process. It is a flowchart which shows an example of the procedure of an evaluation ranking preparation process. It is a flowchart which shows an example of the procedure of an adjustment value calculation process. It is a flowchart which shows an example of the procedure of an evaluation ranking change process. It is a block diagram which shows the hardware structural example of the evaluation ranking search system regarding goods or services. It is explanatory drawing which shows an example of the screen layout of a movie ranking search screen.

  An evaluation rank creating apparatus according to the present embodiment will be described with reference to the drawings. The evaluation rank creating apparatus according to the present embodiment creates an evaluation ranking of products and services based on subjective evaluation of an unspecified number of people. The goods here include tangible and intangibles. The product is any goods such as goods, movable property, real estate, intangibles, such as movie storage media, novels, comic books, game program storage media, music storage media, financial institution products, novelty products, Includes computer programs. The service here is a work by labor or the like, and includes, for example, a movie screening service in a theater, a computer program download service, and a music data download service. Services include services provided at restaurants, ride services on vehicles provided at leisure facilities, services provided at financial institutions, services provided at travel agencies, and the like. In the following, a movie storage medium is cited as an example of a product, and the evaluation order creating device will be described.

Embodiment 1
FIG. 1 is a block diagram illustrating a hardware configuration example of the evaluation rank creating apparatus 10. The evaluation rank creating apparatus 10 is, for example, a general-purpose computer, a workstation, a desktop PC (personal computer), or the like.
The evaluation rank creating apparatus 10 includes a CPU (Central Processing Unit) (collecting unit, dividing unit, counting unit, generating unit, creating unit, changing unit) 11. The evaluation rank creating apparatus 10 includes a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a hard disk 14, a disk drive 15, a display unit 16, an operation unit (accepting unit) 17, a timer 18, and a communication unit ( Reception part) 19 is included.

  The CPU 11 is a processor for executing an instruction set described in a software program. The CPU 11 controls each component of the evaluation rank creating apparatus 10. The CPU 11 reads the program 1P stored in the hard disk 14 to the RAM 13 and executes the program 1P read to the RAM 13.

  The ROM 12 is, for example, a nonvolatile semiconductor memory or a read-only storage medium other than the semiconductor memory. The ROM 12 stores a BIOS (Basic Input / Output System), firmware, and the like that are executed by the CPU 11 when the evaluation rank creating apparatus 10 is activated.

  The RAM 13 is a main storage device. The RAM 13 is, for example, an SRAM or a DRAM, and temporarily stores work variables, data, and the like necessary in the course of processing executed by the CPU 11. Note that a flash memory, a memory card, or the like may be used instead of the RAM 13.

  The hard disk 14 is an auxiliary storage device. The hard disk 14 may be replaced with a flash memory capable of storing a large amount of information or an optical disk 1d such as a CD (Compact Disc), a DVD (Digital Versatile Disc), or a BD (Blu-ray Disc, registered trademark). The hard disk 14 stores a program 1P executed by the CPU 11, various tables, and various files. The various tables include a product service table 1T, a review table 2T, a search keyword table (storage unit) 3T, a connected phrase table (connected phrase storage unit) 4T, a review-specific score table 5T, and a ranking table 6T. Details of these tables will be described later. The various files include a dependency relationship weight value file 1F. Details of the dependency relationship weight value file 1F will be described later.

  The hard disk 14 may be mounted inside the evaluation rank creating apparatus 10 or may be placed outside the evaluation rank creating apparatus 10. For example, various tables stored in the hard disk 14 may be stored in an external database server. In such a case, the evaluation rank creating apparatus 10 accesses an external database server via the communication unit 19 and acquires the contents of various tables. Further, the evaluation rank creating apparatus 10 accesses an external database server via the communication unit 19 and writes information in various tables.

  The disk drive 15 is an auxiliary storage device. The disk drive 15 reads information from an optical disk 1d such as a CD, DVD, or BD that is an external storage medium, and records the information on the optical disk 1d.

  The display unit 16 is a display device that displays an image. The display part 16 has screens, such as a liquid crystal display, an organic EL (Electro Luminescence) display, a CRT (Cathode Ray Tube) display, for example, and displays the various information which concerns on the program 1P according to the instruction | indication from CPU11.

  The operation unit 17 is an input device such as a power switch, an operation button for ejecting the optical disk 1d from the disk drive 15, and a keyboard and a mouse for a user to perform various inputs. The operation unit 17 generates an input signal based on an operation by the user. The generated input signal is transmitted to the CPU 11 via the bus 1b.

  The timer 18 is a device that counts a predetermined time by counting clocks. The timer 18 outputs the time measurement result to the CPU 11.

  The communication unit 19 is a wired or wireless communication modem, a LAN (Local Area Network) card, a router, a USB (Universal Serial Bus) terminal, a connection connector, or the like. The communication unit 19 is connected to a network 1N such as a LAN, a WAN (Wide Area Network), a telephone line, a satellite communication line, and the Internet.

  Note that the CPU 11 may read the program 1P from the optical disc 1d via the disc drive 15. The CPU 11 may read the program 1P from an external information processing apparatus or storage device via the communication unit 19. Further, a semiconductor memory 1m such as a flash memory storing the program 1P may be mounted in the evaluation rank creating apparatus 10.

Next, various tables stored in the hard disk 14 and the dependency relationship weight value file 1F will be described.
FIG. 2 is an explanatory diagram showing an example of a record layout of the product service table 1T. The product / service table 1T is a master table for managing the attributes of products and services. FIG. 2 illustrates the case where the product is a movie storage medium.

  The product service table 1T includes columns of product service ID, product service name, production year, production country, director, and performer. The product service ID is an identification symbol for identifying the product or service. The product service ID is, for example, the product number of the movie optical disc 1d. The product / service name is the name of a product or service, and is the name of a movie in FIG. The production year is the year the movie was produced. The country of production is the country where the movie was produced. The director is the name of the person who directed the movie. The performer is the name of the actor who appeared in the movie. Of course, the merchandise service table 1T may further include a column for storing other attributes relating to the movie.

  FIG. 3 is an explanatory diagram showing an example of a record layout of the review table 2T. The review table 2T is a table that stores a document by subjective evaluation of a movie written by an unspecified number of people on a web page. The program 1P includes a crawler that is a program that periodically acquires documents and images on the cloud and automatically creates a database. The CPU 11 collects reviews written about the movie corresponding to the product service name stored in the product service table 1T from the cloud, and writes the collected reviews to the review table 2T.

  The review table 2T includes columns of product service ID, review ID, written document, and URL (Uniform Resource Locator). The merchandise service ID is the same as the merchandise service ID in the merchandise service table 1T, and corresponds to the name of the movie described on the web page where the CPU 11 has collected the reviews. The review ID is an identification symbol that identifies the review. The written document is a review character string written by an unspecified number of people. The URL is the URL of the target website for which the review has been collected. The review table 2T may include a column that stores other attributes (for example, name, nickname, age, gender, etc.) of the person who wrote the review.

  FIG. 4 is an explanatory diagram showing an example of a record layout of the search keyword table 3T. The search keyword table 3T is a table that stores emotion expression words / phrases associated with scores for each similar word / phrase. Here, the emotion expression phrase is a phrase used when a person expresses a product or service with “feelings”, “taste”, and the like. In addition to those enumerated in FIG. 4, for example, “crying”, “Yandere”, “Yuruyuru”, “squatting”, “exciting”, “exciting”, “fantasy”, “proactive” And the like.

  The search keyword table 3T includes columns of a keyword, a similar word 1, a similar word 2, a similar word 3, a similar word 4, and a similar word 5. The keyword is one of emotion expression phrases. Similar words 1 to 5 are emotion expression phrases similar to the keyword. Of course, the search keyword table 3T may further include a column for storing other similar words.

  The search keyword table 3T includes columns of scores corresponding to the keyword, the similar word 1, the similar word 2, the similar word 3, the similar word 4, and the similar word 5, respectively. The score is a numerical value that serves as a basis for evaluating a product or service based on an emotion expression phrase. In the example of FIG. 4, the score is an integer, but the score may be a real number. Not only the search keyword table 3T but also a numerical value handled by the evaluation rank creating apparatus 10 for creating an evaluation ranking related to products or services may be a real number.

  FIG. 5 is an explanatory diagram showing an example of the record layout of the connected word / phrase table 4T. The connected word / phrase table 4T is a table that stores words associated with emotion expression words / phrases (keywords or similar words) in the search keyword table 3T in association with points. Here, the phrase connected to the emotion expression phrase is a phrase having a dependency relationship with the emotion expression phrase, for example, an auxiliary verb, an adjective, an adverb or the like that denies or emphasizes the emotion expression phrase.

  The connected phrase table 4T includes columns of keyword-similar word, connected phrase, and score. The keyword-similar words are keywords, similar words 1 to 5 in the search keyword table 3T. A connected phrase is a phrase connected to an emotional expression phrase. The score is a numerical value for adjusting the numerical value corresponding to the keyword or similar words 1 to 5 in the search keyword table 3T when the emotion expression word and the word connected to the emotion expression word are included in the review. is there. For example, when the dependency relationship between the emotion expression phrase and the connected phrase is negative, a negative numerical value is set in the score of the connected phrase table 4T. When the dependency relationship between the emotion expression phrase and the connected phrase is positive emphasis, a positive numerical value is set to the score of the connected phrase table 4T. In addition, when the connected phrase is a phrase that blurs an emotion expression phrase or an ambiguous phrase, a negative numerical value may be set for the score.

  FIG. 6 is an explanatory diagram showing an example of a record layout of the review-specific score table 5T. The review-specific score table 5 </ b> T is a table that stores numerical values that are tabulated based on keywords and similar words included in reviews for each product service and each review. Below, the numerical value totaled for every product service and every review is called a score.

  The review-specific score table 5T includes columns of product service ID, review ID, keyword, and score. The product service ID is the same as the product service ID in the product service table 1T. The review ID is the same as the review ID of the review table 2T. The keyword is the same as the keyword in the search keyword table 3T. The score is a numerical value calculated based on keywords and similar words included in the review.

  FIG. 7 is an explanatory diagram showing an example of the record layout of the ranking table 6T. The ranking table is a table that stores the evaluation ranking of products or services by product or service.

  The ranking table 6T includes columns of rank, product service name, keyword, total number of reviews, and total score. The rank is a number indicating the descending order of the overall score. The product / service name is the name of the product or service to be evaluated. The keyword is a keyword in the search keyword table 3T. The total number of reviews is the total number of reviews that evaluated one product or service. The total score is a numerical value calculated based on the total score of each review for each product service and the total number of reviews.

FIG. 8 is an explanatory diagram showing an example of the contents of the dependency relationship weight value file 1F. The dependency relationship weight value file 1F is a file that stores a weight value obtained by quantifying the strength of the dependency relationship recognized between words included in the review. The weight value is generated by the evaluation ranking device 10 based on a certain rule between words. For example, the weight value is set to a smaller value as the distance in the review between words in the dependency relationship is longer.
FIG. 8 exemplifies the weight value generated for the review example “This movie is a comedy and cannot be cryed”.

  The evaluation rank creating apparatus 10 creates a dependency relationship weight value file 1F and stores it in the hard disk 14 each time a process for creating an evaluation ranking related to products or services is executed. That is, the dependency relationship weight value file 1F is a working file. The dependency relationship weight value file 1F may be stored in the RAM 13 as an array, for example.

  FIG. 9 is a functional block diagram illustrating a functional configuration example of the evaluation rank creating apparatus 10. Each functional unit of the evaluation rank creating apparatus 10 is realized by a program 1P stored in the hard disk 14 and hardware resources such as the CPU 11 and the RAM 13 operating in cooperation.

The evaluation rank creating apparatus 10 includes a storage unit 20, a collecting unit 30, a dividing unit 40, a calculating unit 50, a generating unit 60, a totaling unit 70, a creating unit 80, a receiving unit 90, and a changing unit 100.
The storage unit 20 is a functional unit corresponding to the hardware hard disk 14 and the RAM 13. The memory | storage part 20 has memorize | stored various tables required in order to produce the evaluation ranking of goods or services. The storage unit 20 is used by the CPU 11 to temporarily store work data.

  The collection unit 30 is a functional unit corresponding to a crawler that collects documents relating to evaluation of products or services from the cloud. The collection unit 30 refers to the product / service table 1T, collects reviews written by an unspecified number of people from a web page including the name of the product or service, and uses the collected reviews as written documents in the review table 2T. Write. In the present embodiment, the review is a document related to a critique of a movie.

  The dividing unit 40 is a functional unit that decomposes the reviews collected by the collecting unit 30 using morphological analysis. The dividing unit 40 reads a review from a document written in the review table 2T for each product or service, and divides the read document into parts of speech. The dividing unit 40 stores the part of speech, that is, the review divided into words in the storage unit 20.

  The calculation unit 50 is a functional unit that calculates a review score for each product or service and for each review. The calculating unit 50 searches the keywords and the similar words 1 to 5 in the search keyword table 3T from the review words divided by the dividing unit 40. When the keyword or similar words 1 to 5 in the search keyword table 3T are searched from the word of the review, the calculation unit 50 extracts the score of the search keyword table 3T corresponding to the searched keyword or similar words 1 to 5. The calculation unit 50 uses the extracted score to calculate a score for each product or service and for each review.

Next, a process in which the calculation unit 50 calculates the review-specific score will be described by exemplifying specific reviews, keywords, and similar words.
The review now assumes that this movie was very crying and tearing. It is assumed that the keyword for evaluating this review is “cry” and the similar word of the keyword is “tears”. Further, in the search keyword table 3T, the score of the keyword “can cry” and the similar word “tears” are 3 points and 2 points, respectively.
The dividing unit 40 divides the reviews into 10 parts of speech of “this”, “movie”, “ha”, “very”, “can cry”, “te”, “tears”, “ga”, “out”, “ta”. Divide into

  The calculation unit 50 calculates the score for each review by using the following formula (1) in consideration of the ratio of emotion expression phrases in the review.

  First, the calculation unit 50 calculates the total score for each part of speech, which is a numerator on the right side of Equation (1). However, the calculation unit 50 adjusts the score corresponding to the keyword using the weight value generated by the generation unit 60 prior to calculating the total score for each part of speech. In the case of the above example, the calculation unit 50 calculates the weight value generated by the generation unit 60 for three points “cry” before adding three points (“cry”) and two points (“tears”). Use to adjust. Therefore, the generation unit 60 will be described before the calculation process of the total score for each part of speech by the calculation unit 50 is described.

  The generation unit 60 is a functional unit that generates the dependency relationship weight value file 1F. The generating unit 60 performs dependency analysis on the review words divided by the dividing unit 40 and generates a weight value for each combination of words and words according to a certain rule. The generation unit 60 stores the generated weight value for each combination in the storage unit 20 as a dependency relationship weight value file 1F.

The calculation unit 50 converts the score corresponding to the keyword into an adjustment value adjusted by the weight value. Specifically, the adjustment value = the number of points in the search keyword table 3T × the number of points in the connected phrase table 4T × the weight value of the dependency relationship weight value file 1F.
For example, in the connected word / phrase table 4T, it is assumed that the keyword-similar word = “cry”, the connected word = “very”, and the score = 1.2 (see FIG. 5). In the dependency relationship weight value file 1F, the weight value for the combination of “very” and “cry” is 5 points. In this case, the adjustment value of the keyword “can cry” = 3 points (“cry”) × 1.2 points (“very”) × 5 points (weight value for the combination of “very” and “cry”) = 18 points Become.
The adjustment value may be a positive value or a negative value depending on the score of the connected word / phrase table 4T. The above illustrates the case where the adjustment value is a positive value.

The calculation unit 50 adds the score corresponding to the similar word of the keyword to the adjustment value corresponding to the keyword to obtain the total score for each part of speech. The total score for each part of speech in the above example is shown below.
Total score by part of speech = 18 points (adjusted value for “crying”) + 2 points (“tears”) = 20 points

Since the total number of parts of speech included in the formula (1) is 10,
Total number of parts in review + 1 = 10 + 1 = 11
It is.
Therefore, from equation (1),
Review score = 20 / ln11 = 8.340648 points.

  The calculation unit 50 writes the calculated score for each review together with the product service ID, the review ID, and the keyword in the score column of the score table for each review 5T.

  The totaling unit 70 is a functional unit that totals the reviews-specific scores calculated by the calculation unit 50 for each product or service. The totaling unit 70 considers the total number of reviews and totals the product or service-specific scores using the following formula (2).

  The total number of reviews in equation (2) is the total number of reviews that have evaluated the same product or service, and is the same as the total number of reviews in the ranking table 6T. The total number of reviews for each product or service is a numerical value obtained by adding up the scores of each review that evaluated the same product or service. The totaling unit 70 totals the scores having the same product / service ID from the review-specific score table 5 </ b> T to obtain the total number of reviews for each product or service. The totaling unit 70 writes the product or service-specific score obtained using Expression (2) in the overall score of the ranking table 6T. The totaling unit 70 also writes the corresponding data in the columns of the ranking table 6T other than the ranking (product service name, keyword, and total number of reviews).

  The creation unit 80 is a functional unit that creates an evaluation ranking for each product or service. Specifically, the creation unit 80 is a functional unit that rearranges the records of the ranking table 6T in descending order of the total score for each product service name. The creation unit 80 writes the numbers corresponding to the descending order of the total score in the rank sequence of the records rearranged in the ranking table 6T.

  The reception unit 90 is a functional unit corresponding to the hardware operation unit 17 or the communication unit 19. The reception unit 90 receives an evaluation ranking related to a result of an evaluation questionnaire performed on a product or service. The questionnaire here includes a street questionnaire, a web-based questionnaire, a telephone questionnaire, a mail-based questionnaire, a store questionnaire, and the like.

  The changing unit 100 compares the evaluation ranking received by the receiving unit 90 with the evaluation ranking created by the creation unit 80 for the same product or service, and based on the comparison result, the evaluation ranking created by the creation unit 80 It is a functional part to be changed. One of the comparison processes performed by the changing unit 100 is, for example, calculating a correlation coefficient between an evaluation ranking based on a questionnaire and an evaluation ranking created by the creating unit 80. The correlation coefficient here is, for example, a general correlation coefficient, Spearman's rank correlation coefficient, Kendall's rank correlation coefficient, or the like.

  The change unit 100 does not change the evaluation ranking created by the creation unit 80 when the calculated correlation coefficient is a certain value (for example, 0.7) or more. On the other hand, when the calculated correlation coefficient is equal to or less than a predetermined value, the changing unit 100 compares the evaluation ranking created by the creating unit 80 with the evaluation ranking based on the questionnaire, and determines the evaluation ranking created by the creating unit 80, for example. Change from a higher level.

  For example, the changing unit 100 changes the evaluation ranking created by the creating unit 80 from the higher rank where the difference from the evaluation ranking based on the questionnaire is generated one by one in a direction approaching the evaluation ranking based on the questionnaire. The changing unit 100 calculates again the correlation coefficient between the changed evaluation ranking and the evaluation ranking based on the questionnaire, and performs the above change and the calculation of the correlation coefficient until the calculated correlation coefficient is equal to or greater than a certain value. repeat. When the calculated correlation coefficient becomes a certain value or more, the changing unit 100 stops changing the evaluation ranking, and determines the evaluation ranking after the change as the evaluation ranking after the change.

Next, the operation of the evaluation order creating device 10 will be described.
10 and 11 are flowcharts illustrating an example of the procedure of the evaluation ranking creation process. The CPU 11 refers to the merchandise / service table 1T and collects documents related to the evaluation of the merchandise or service from the cloud (step S101). The CPU 11 writes the collected document in the review table 2T (step S102). The CPU 11 determines whether or not the number of records in the review table 2T is greater than or equal to a certain value (step S103).

  If the CPU 11 determines that the number of records in the review table 2T is not equal to or greater than a certain value (step S103: NO), the process returns to step S101. If the CPU 11 determines that the number of records in the review table 2T is equal to or greater than a certain value (step S103: YES), the process proceeds to step S104. CPU11 repeats the process of step S101-step S103, and increases the number of records of the review table 2T to a fixed value or more. Thereby, it is possible to guarantee the statistical reliability of the number of reviews (number of samples) that is the basis of the created evaluation ranking.

  CPU11 receives goods service ID for evaluating a review, and the keyword which concerns on an emotion expression phrase (step S104). The CPU 11 reads one review related to an unprocessed written document corresponding to the received product service ID from the top of the review table 2T (step S105). The CPU 11 decomposes the part of the read review by morphological analysis (step S106). The CPU 11 reads the similar words 1 to 5 corresponding to the keyword received in step S104 from the search keyword table 3T (step S107).

  The CPU 11 searches the received keyword and the read similar words 1 to 5 from the part of speech of the part-of-speech decomposed review, and determines whether the search is successful (step S108). If the CPU 11 determines that the search has not been successful (step S108: NO), the process proceeds to step S114. If the CPU 11 determines that the search is successful (step S108: YES), the CPU 11 extracts the score corresponding to the keyword or similar word from the search keyword table 3T (step S109). CPU11 performs an adjustment value calculation process (step S110). The adjustment value calculation process is a process of calculating an adjustment value corresponding to a keyword in order to calculate the total score for each part of speech.

  The CPU 11 calculates the total score for each part of speech by adding the score corresponding to the keyword similar word extracted in step S109 to the adjustment value corresponding to the keyword calculated in step S110 (step S111). The CPU 11 calculates a review-specific score based on the total score for each part of speech, the total number of parts of speech within the review, and the formula (1) (step S112). The CPU 11 writes the adjusted score by review in the score column of the review-specific score table 5T (step S113).

  The CPU 11 determines whether or not all the records in the review table 2T have been processed (step S114). If the CPU 11 determines that all the records in the review table 2T have not been processed (step S114: NO), the process returns to step S105. If the CPU 11 determines that all the records in the review table 2T have been processed (step S114: YES), the CPU 11 calculates a product or service-specific score based on the contents of the review-specific score table 5T, the total number of reviews, and formula (2). (Step S115).

  The CPU 11 writes the calculated product or service-specific score in the total score column of the ranking table 6T (step S116). In step S116, the CPU 11 writes corresponding data in columns other than the rank. The CPU 11 rearranges the records of the ranking table 6T in descending order of the total score (step S117). The CPU 11 writes numbers corresponding to the descending order of the total score in the rank sequence of the records rearranged in the ranking table 6T (step S118), and ends the process.

FIG. 12 is a flowchart illustrating an example of the procedure of the adjustment value calculation process.
CPU11 produces | generates the weight value according to the intensity | strength of dependency relation for every combination of the words of the review which carried out the part-of-speech decomposition (step S201). The CPU 11 stores the generated weight value for each combination in the hard disk 14 as a dependency relationship weight value file 1F (step S202).

  The CPU 11 refers to the connected phrase table 4T to determine whether or not the keyword accepted in step S104 and the phrase connected to the keyword are included in the review word partly decomposed (step S203). CPU11 complete | finishes a process, when it determines with the keyword received by step S104 and the phrase connected with the said keyword not being included in the word of the review which decomposed | disassembled the part of speech (step S203: NO). If the CPU 11 determines that the keyword received in step S104 and the phrase connected to the keyword are included in the word of review that has undergone part-of-speech decomposition (step S203: YES), the CPU 11 executes the following process. That is, the CPU 11 extracts, from the connected phrase table 4T, the score corresponding to the keyword received in step S104 and the phrase connected to the keyword (step S204).

  The CPU 11 reads the weight value corresponding to the combination of the keyword received in step S104 and the phrase connected to the keyword from the dependency relationship weight value file 1F (step S205). The CPU 11 calculates the product of the score corresponding to the keyword extracted in step S109, the score extracted in step S204, the score corresponding to the word connected to the keyword, and the weight value read in step 205 (step S206). . The CPU 11 determines the calculated product as an adjustment value (step S207), and ends the process.

FIG. 13 is a flowchart illustrating an example of the procedure of the evaluation ranking changing process.
CPU11 receives the evaluation ranking which concerns on the result of the evaluation questionnaire implemented about goods or service from the operation part 17 or the communication part 19 (step S301). The CPU 11 calculates a correlation coefficient between the evaluation ranking created in step S118 or the evaluation ranking changed in step S304 and the evaluation ranking related to the accepted questionnaire (step S302). In step S302, in the first loop (steps S302 to S304), the CPU 11 calculates a correlation coefficient between the evaluation ranking created in step S118 and the evaluation ranking related to the accepted questionnaire. In step S302, in the second and subsequent loops (steps S302 to S304), the CPU 11 calculates a correlation coefficient between the evaluation ranking changed in step S304 and the evaluation ranking related to the accepted questionnaire.

  The CPU 11 determines whether or not the calculated correlation coefficient is greater than or equal to a certain value (step S303). CPU11 advances a process to step S305, when it determines with the calculated correlation coefficient being more than a fixed value (step S303: YES). If the CPU 11 determines that the calculated correlation coefficient is not equal to or greater than a certain value (step S303: NO), the CPU 11 selects a part of the evaluation ranking created by itself, which is one of the targets for which the correlation coefficient has been calculated in step S302. Change from the upper level (step S304), and return the process to step S302.

  The CPU 11 exits the loop (steps S302 to S304), determines the evaluation ranking that has proceeded from step S303 to step S305 as the changed evaluation ranking (step S305), and ends the process.

In step S303, the value to be compared with the correlation coefficient is constant. However, the value to be compared with the correlation coefficient in step S303 may not be constant. For example, the value compared with the correlation coefficient in the second loop may be smaller than the value compared with the correlation coefficient in the first loop. The value compared with the correlation coefficient in the third loop may be smaller than the value compared with the correlation coefficient in the second loop. Thereby, the change degree of the evaluation ranking changed by the evaluation rank creating apparatus 10 can be reduced, and the time required for the evaluation ranking changing process can be shortened.
The evaluation ranking according to the questionnaire and the evaluation ranking based on the reviews collected from the cloud have different properties reflecting the difference in the original data. Therefore, rather than bringing the evaluation ranking based on reviews collected from the cloud closer to the evaluation ranking related to the questionnaire to the dark clouds, the nature of the evaluation ranking based on the reviews collected from the cloud is kept within a certain range by suppressing the degree of change. To help.

  In the above-described example, the evaluation rank creating apparatus 10 calculates the score for each review by adjusting the score corresponding to the keyword. However, the evaluation rank creating apparatus 10 may not adjust the score corresponding to the keyword. As a result, the load on the evaluation rank creating apparatus 10 can be reduced by the amount that eliminates the process of generating the adjustment value.

  In the above-described example, the score value of the search keyword table 3T × the score of the connected word / phrase table 4T × the weight value of the dependency relationship weight value file 1F is used as the keyword adjustment value. However, the adjustment value may be the score of the search keyword table 3T × the score of the connected phrase table 4T or the score of the search keyword table 3T × the weight value of the dependency relationship weight value file 1F. As a result, the load on the evaluation rank creating apparatus 10 can be reduced by the amount of lightening the process of generating the adjustment value.

  In the above-described example, the evaluation rank creating apparatus 10 decomposes the reviews into parts of speech and creates an evaluation ranking. However, the evaluation rank creating apparatus 10 may divide the review into phrases. In such a case, points are set for each phrase in the search keyword table 3T and the connected phrase table 4T. In the dependency relationship weight value file 1F, a weight value corresponding to the strength of the dependency relationship in phrase units is set.

  In the above example, the score of the search keyword table 3T is set to a different score for each keyword and similar word. However, the points in the search keyword table 3T may all be the same. In such a case, the total score for each part of speech before adjusting the score corresponding to the keyword is proportional to the number of keywords or similar words included in the review.

The evaluation rank creating apparatus 10 may include an anaphoric analysis means for complementing the subject and the object omitted in the review. If the omitted phrase is an emotional expression phrase or a phrase that leads to an emotional expression phrase, creating a rating ranking after supplementing the omitted phrase to the review can create a more accurate evaluation ranking. It becomes possible.
For example, in Japanese, subjects, objects, etc. are frequently omitted. When the evaluation rank creating apparatus 10 performs morphological analysis on a review that includes omission of a phrase, the abbreviation is not included in the part of speech that has been decomposed. Therefore, the evaluation rank creating apparatus 10 supplements the review with a noun phrase (zero pronoun) omitted by the known anaphora analysis.

According to the evaluation rank creating apparatus 10, it is possible to create a highly reliable evaluation ranking regarding products or services.
Since the evaluation rank creating apparatus 10 creates an evaluation ranking related to products or services based on a large number of reviews collected from the cloud, the created evaluation ranking has high statistical reliability and objectivity. In addition, since there is no limit to the number of emotion expression phrases that can be registered in the search keyword table 3T, the evaluation rank creating apparatus 10 provides evaluation ranks for products or services that match a wide variety of “feelings” and “taste”. Can do.

  The evaluation rank creating apparatus 10 calculates the score for each review by adjusting the score corresponding to the keyword in consideration of the dependency relationship of the negative word or the emphasized word connected to the emotion expression word. Thereby, the evaluation ranking creating apparatus 10 can create a more detailed evaluation ranking with respect to the product or service.

  The evaluation rank creating apparatus 10 confirms the correlation between the evaluation ranking based on the questionnaire regarding the product or service and the evaluation ranking regarding the product or service created by itself. Thereby, the quality of the evaluation ranking created by the evaluation ranking creation device 10 is secured. On the other hand, when the correlation is low, the evaluation rank creating apparatus 10 can correct the evaluation ranking related to the product or service created by itself.

Embodiment 2
The second embodiment relates to a form in which an evaluation ranking related to products or services created by the evaluation ranking creating apparatus 10 is disclosed on the Internet.
In the second embodiment, the same reference numerals are assigned to the same components as those in the first embodiment, and detailed description thereof is omitted.

  FIG. 14 is a block diagram illustrating a hardware configuration example of the evaluation ranking search system 200 related to products or services. The evaluation ranking search system 200 includes an evaluation ranking creation device 10, a web server 210, and a terminal device 220.

Evaluation rank creating apparatus 10 is the same as evaluation rank creating apparatus 10 in the first embodiment.
The web server 210 is, for example, a general-purpose computer, a workstation, a desktop PC, or the like. The web server 210 is connected to the evaluation rank creating apparatus 10 through a dedicated line 1n. The web server 210 is connected to the Internet 1N.

The terminal device 220 is a desktop PC, notebook PC, tablet PC, smartphone, or the like. The terminal device 220 is connected to the Internet 1N. In FIG. 14, three terminal devices 220 are illustrated as an example, but the number of terminal devices 220 is not limited to three, and may be four or more.
In the following, it is assumed that the product is a movie optical disk 1d.

  A browser is installed in the terminal device 220. The terminal device 220 transmits a keyword and a web page information transmission request to the web server 210. When the web server 210 receives a keyword and a web page information transmission request from the terminal device 220, the web server 210 uses the received keyword and information indicating that the product is a movie optical disk 1 d as the evaluation rank creating apparatus 10. Send to.

  When receiving the keyword and the information indicating that the product is the movie optical disk 1d from the web server 210, the evaluation rank creating apparatus 10 creates an evaluation ranking related to the movie optical disk 1d based on the received keyword. The evaluation rank creating apparatus 10 transmits the created evaluation ranking to the web server 210. The web server 210 generates information on a web page that displays an evaluation ranking related to the optical disc 1 d of the movie, and transmits the generated web page information to the terminal device 220.

  The terminal device 220 receives web page information related to the evaluation ranking related to the optical disc 1 d of the movie from the web server 220. Based on the transmitted information, the terminal device 220 displays a web page related to the evaluation ranking related to the optical disc 1d of the movie on the browser screen.

  FIG. 15 is an explanatory diagram showing an example of the screen layout of the movie ranking search screen 1f. The movie ranking search screen 1f is an example of a web page displayed on the browser screen of the terminal device 220 when a transmission request is made from the terminal device 220 to the web server 210 by specifying “don't return” as a keyword. As in the example illustrated in FIG. 15, the keyword that is an emotion expression phrase may include a phrase related to the story development of a story appealing to human sensitivity.

The movie ranking search screen 1f includes a text box 1txt, a search button 1btn, an evaluation ranking list 1lst, a forward button 2btn, and a return button 3btn.
A text box 1txt is a button for inputting a keyword related to an emotion expression phrase. The search button 1btn is a button for setting a keyword input in the text box 1txt as a parameter and transmitting a web page information transmission request to the web server 210.

  The evaluation ranking list 1lst is a screen area that displays a list of rankings of the optical disc 1d of the movie searched based on the keyword input in the text box 1txt. In the evaluation ranking list 1lst, the searched movies are arranged in order of ascending order from left to right. The information of each movie displayed in the evaluation ranking list 1lst is, for example, the image of the package of the optical disc 1d, the ranking, the overall score, the title of the movie, the name of the representative reviewer, and the score for each review regarding the review of the representative reviewer. . The information of each movie displayed in the evaluation ranking list 11st is, for example, a part of reviews (comments) of typical reviewers, the total number of reviews (number of reviews), and the like.

  The forward button 2btn is a button for displaying the ranking of lower-order movies in the evaluation ranking list 1lst. The return button 3btn is a button for displaying the ranking of a higher-order movie in the evaluation ranking list 1lst.

  In the above-described example, the evaluation rank creating apparatus 10 creates an evaluation ranking related to products or services. However, the evaluation rank creating apparatus 10 does not have to create the evaluation ranking by executing the processes from adding the scores for each review and calculating the total score. In other words, the evaluation rank creating apparatus 10 only needs to execute processing until creating a record relating to the product service name, keyword, total number of reviews, and total score in the ranking table 6T, and rearranging the records in the ranking table 6T. It is not necessary to execute. Further, the evaluation rank creating apparatus 10 may not write the rank in the ranking table 6T.

  In such a case, the web server 210 of the evaluation ranking search system 200 causes the terminal device 220 to download data related to the product or service total score collected by the evaluation rank creating apparatus 10. The data is, for example, a spreadsheet software file. The user causes the spreadsheet software installed in the terminal device 220 to read the data related to the product or service total score downloaded from the web server 210 to the terminal device 220. Then, the user causes the terminal device 210 to create a product or service evaluation ranking using the sorting function of the spreadsheet software.

The evaluation rank creating apparatus 10 may have means for deleting the record of the review table 2T according to a certain rule.
The evaluation rank creating apparatus 10 circulates a web server on the cloud and collects reviews regarding products or services. For this reason, the number of records in the review table 2T increases with the passage of time, leading to a decrease in the usable storage capacity of the hard disk 14 and an increase in the execution time of the evaluation ranking creation process. Therefore, the evaluation rank creating apparatus 10 eliminates the above-described adverse effects by deleting records in the review table 2T according to a certain rule.
The certain rule here is, for example, deletion of records older than a certain time from the latest. Old reviews tend to depart from the preferences of the latest users, and the evaluation ranking based on old reviews is not necessarily the evaluation ranking desired by the user. Based on the time stamp of the record, the evaluation rank creating apparatus 10 deletes a record older than a certain time from the latest from the review table 2T.

The evaluation rank creating apparatus 10 may have means for deleting records in the search keyword table 3T according to a certain rule.
Here, the certain rule is to delete records corresponding to keywords that the user has used less frequently than a certain number of times in the evaluation ranking search system 200. In spite of many years of operation, keywords that are hardly or not transmitted from the terminal device 220 to the web server 210 should be deleted from the search keyword table 3T, and new keywords and similar words should be added to the search keyword table 3T. . For this purpose, a table (hereinafter referred to as a keyword history table) for storing the transmission frequency of the keyword transmitted from the terminal device 220 is prepared in the hard disk 14 of the evaluation rank creating apparatus 10. The evaluation rank creating apparatus 10 updates the keyword transmission frequency in the keyword history table based on the keyword transmitted from the web server 210. When the evaluation rank creating apparatus 10 detects a keyword whose keyword transmission frequency is equal to or less than a certain number of times after a certain time has elapsed, the evaluation rank creating apparatus 10 deletes the record of the search keyword table 3T corresponding to the keyword from the search keyword table 3T. .
Further, the evaluation rank creating apparatus 10 may delete the record of the connected phrase table 4T corresponding to the keyword and / or similar word from the connected phrase table 4T.

  According to the evaluation ranking search system 200, the user can display the evaluation ranking related to the product or service created by the evaluation rank creating device 10 on the screen of the terminal device 220 used by the user via the web server 210.

  The disclosed embodiments are to be considered in all respects as illustrative and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

  Further, the technical features (components) described in each embodiment can be combined with each other, and a new technical feature can be formed by combining them.

10 evaluation rank creating device 11 CPU (collecting unit, dividing unit, totaling unit, generating unit, creating unit, changing unit)
13 RAM
14 Hard disk 17 Operation part (reception part)
19 Communication Department (Reception Department)
3T search keyword table (storage)
4T connection phrase table (connection phrase storage unit)
1P program 1F Dependency weight file

Claims (9)

An evaluation and aggregation device that aggregates numerical values related to evaluation of goods or services based on documents collected from a website,
A collection unit that collects from the website a document relating to the evaluation of the goods or services written on the web page describing the names of the goods or services;
A dividing unit for dividing the document collected by the collecting unit into words;
A storage unit that pre-stores a word that matches and resembles the emotion expression word and a numerical value corresponding to the word respectively in advance,
A tabulation unit that counts numerical values corresponding to the words when the words divided by the dividing unit include words that match or are similar to the emotion expression words based on the contents of the storage unit, To evaluate and aggregate. An emotional expression phrase stored in the storage unit, a phrase connected to the emotional expression phrase and a connected phrase storage unit stored in advance in association with numerical values;
The aggregation unit, when the phrase divided by the dividing unit includes the emotion expression phrase and the phrase connected to the emotion expression phrase, numerical values corresponding to the emotion expression phrase and the connected phrase stored in the connection phrase storage unit The evaluation totaling apparatus according to claim 1, wherein numerical values corresponding to emotion expression words and phrases stored in the storage unit are adjusted and totaled. For a combination of words divided by the dividing unit, a generation unit that generates a weight value according to the dependency relationship of words related to the combination,
When the words divided by the dividing unit include the emotion expression words and phrases connected to the emotion expression words, the counting unit includes a numerical value corresponding to the emotion expression words and connected words stored in the connection word storage unit; The weight value generated by the generation unit for the combination of the emotion expression phrase and the connected phrase is adjusted, and the numerical value corresponding to the emotion expression phrase stored in the storage unit is adjusted and aggregated. The evaluation tabulation apparatus according to claim 2, wherein The evaluation totaling apparatus according to claim 2 or 3, wherein a phrase connected to an emotion expression phrase stored in the connection phrase storage unit is a negative phrase or an emphasized phrase for the emotion expression phrase. The evaluation tabulation apparatus according to any one of claims 1 to 4,
An evaluation rank creating apparatus comprising: a creation section that creates an evaluation rank of the product or service based on numerical values collected by the counting section of the evaluation aggregation apparatus. 6. The evaluation rank creating apparatus according to claim 5, wherein the creation section creates the evaluation rank based on the number of documents collected for each product or service by the collection section. A reception unit for receiving evaluation ranks of products or services based on a questionnaire;
The evaluation unit received by the reception unit and the evaluation rank created by the creation unit are compared, and a change unit that changes the evaluation rank created by the creation unit is provided. The evaluation ranking creation device described in 1. An evaluation tabulation method for tabulating numerical values related to evaluation of products or services based on documents collected from a website,
Collect from the website documents related to the evaluation of the goods or services written on the web page containing the name of the goods or services;
Divide the collected documents into words,
Based on the contents of the storage unit preliminarily stored in association with numerical values corresponding to the respective phrases that match and are similar to the emotion expression words, the divided phrases include words that match or are similar to the emotion expression words If it is, an evaluation tabulation method characterized in that a numerical value corresponding to the word is tabulated. A program for causing a computer having a storage unit to execute a process of aggregating numerical values related to evaluation of goods or services based on documents collected from a website,
Collect from the website documents related to the evaluation of the goods or services written on the web page containing the name of the goods or services;
Divide the collected documents into words,
Based on the contents of the storage unit that pre-stores the words that match and are similar to the emotion expression words and the numerical values that respectively correspond to the words, the words that match or are similar to the emotion expression words are included in the divided words If included, a program that causes a computer to execute a process of aggregating numerical values corresponding to the words.

Images